Forecasting

How to tailor forecasts to the unique characteristics of individual markets.

How to tailor forecasts to the unique characteristics of individual markets.

Although critics of the EU see it as progressing to a kind of monolithic super-state, its healthcare systems are highly diverse. Lee Blansett, senior VP at Kantar Health, sees the EU as kind of laboratory in which different reimbursement models are piloted, some of which may be exported to the US. But there is two-way traffic, with US ideason restricting use of drugs, for exampleexported to the EU.

Essentially, EU healthcare systems fall into two groups. The decentralized systems of Spain and Italy contrast with the centralization in the UK, France and, increasingly, Germany. Thus, forecasting in Spain and Italy is a lot more complex, says Blansett.

Centralized and decentralized systems

Notwithstanding the diversity of systems, overall the cost containment measures implemented throughout Europe are having the desired effect. Blansett highlights Italy, where the pharmaceutical proportion of total healthcare spending will fall, by law, in 2010 from 14% to 13.3%, a saving of around 800 million compared with 2008. Doctors and hospitals are very difficult budgets to cut, observes Blansett; manufacturers are much easier to target.

The centralized system in France has two main factors, the launch price and how quickly that price comes up for renegotiation, explains Blansett. These factors are interdependentthe higher the price at launch, the shorter the price protection. However, it can be more complex than that, as pricing is linked to expected sales volume. If the latter differs from the estimate, more price concessions could be demanded. Forecasting in France is challenging and it is going to get worse, says Blansett.

Outcome-based measures

The trend is for healthcare decisions to be increasingly based on value, and this will include impacts on many areas. Health technology assessments are growing rapidly, using the familiar quality adjusted life year (QALY) but also newer measures, such as incremental cost effectiveness ratio (ICER) and the efficiency frontier from IQWIG in Germany. All these are, of course, outcome-based measures, and measuring a change in outcome is a complete witch's brew, according to Blansett. In addition, in Europe, there are not only multiple countries but, in the decentralized systems, multiple regions, too.

However, there is evidence of US models being taken up in Europe. Contracting, a familiar concept to US forecasters, is already in place in Germany, with interest coming from Spain and Italy. The UK has not taken up contracting, but has embraced risk sharing. There are also sophisticated risk sharing schemes in Italy and Germany.

But Blansett warns that these arrangements can have impacts on forecasts of sales volumes. Complexity is further driven by the science; he cites the fragmentation of lung cancer indications as an example. Genetic testing, in particular, has dramatically narrowed the indications for some drugs, generating what Blansett calls instant orphans.

Monte Carlo simulations

Todd R. Johnson, director of forecasting at Kantar Health, emphasizes that forecasters must think globally and act locally, particularly with regard to market research. Most surveys use the Internet, and in Europe consideration needs to be given to issues like random sampling, language, and local knowledge.

Inaccurate estimation of market share is a risk everywhere, but this may be country specific. For example, in Italy estimates are known to be over-optimistic, but perhaps pessimistic in Germany. So Johnson thinks it is unwise to use the same correction factor across Europe. Likelihood to prescribe is a commonly used analogue, and this is also country-specific. Johnson says uptake curves vary across countries, leading to different times to peak sales.

This is where Monte Carlo simulations can help. The method calculates the probability of deviating outside the limits of forecasts, and is useful where there is more uncertainty, as in the EU. But he emphasizes that Monte Carlo simulations do not obviate the need for good quality forecasting you should reduce uncertainty as far as possible before doing the simulation. There will be different sets of competitors among countries, says Johnson, and it is all too easy to forget about currency fluctuations. His take-home messages: Tailor forecasts to each market's unique characteristics, use country-specific analogues (derived from databases of experience), and use Monte Carlo simulations where appropriate.